| ▲ | _flux 3 months ago |
| Many years ago there used to be a Firefox extension (..or might have even been a Mozilla one..) that would store all the pages I visit. I recall its name was Breadcrumbs but I could be misremembering. Space is cheap, or at least affordable if one would exclude videos, which are probably technically more difficult to archive anyway, but sometimes one remembers having seen content that is never to be found again. I think it would be useful to have just a personal basic search engine on that kind of contents, but possibly a RAG or even a fine tuned LLM would be even cooler. Actually, e.g. Firefox could do that at least for its bookmarks and tabs, though it already does provide the function for tagging bookmarks. And I think there's probably an extension for searching tabs' contents.. |
|
| ▲ | irthomasthomas 3 months ago | parent | next [-] |
| Not identical but I started building a smart bookmark tool that stores the content in vectors and sqlite dB and hosts them in GitHub issues with labels managed by the ai. Check it: https://undecidability.com and code lives at https://github.com/irthomasthomas/label-maker
It's a bit rough but there is a working cli. It uses local jina embeddings model but openai logprobs to determine when to create new labels. |
|
| ▲ | fire_lake 3 months ago | parent | prev | next [-] |
| Given how personal browsing history can be this is a great use case for local LLMs. I would love for Mozilla to deliver on this. |
| |
| ▲ | jumping_frog 3 months ago | parent [-] | | Building personal assistant could be beneficial to Mozilla based on how much we do online. I would like to track changes to my beliefs based on how I came across new information. In future, the AI could automatically shorten paragraphs in essays about topics or terms I am already aware of while keeping new concepts introduced in it full expanded so that I grok them better. |
|
|
| ▲ | TiredOfLife 3 months ago | parent | prev | next [-] |
| The original version of read it later (now Mozilla owned Pocket) had that option. but then removed that option because it went against their commercial interests. |
| |
| ▲ | monkeydust 3 months ago | parent [-] | | Pocket is good. I use it across all my devices, simple and works for me but do wonder if they could or should do more with the data they collect from me which is all the things I really care about. | | |
|
|
| ▲ | gazreese 3 months ago | parent | prev [-] |
| I need this so much, someone please build it ASAP. This would be so useful! |
| |
| ▲ | mehh 3 months ago | parent [-] | | Working on it https://ont.fyi The approach is not to capture all pages you view, rather you can add the pages etc you want in order to reduce the amount of noise/rubbish. It constructs a knowledge graph from these documents, and then a graph rag approach ontop to enable chat. The core graph is based on wikidata, you can have your graphs either private or public if so they are published like those you can see on the site now. Lots to do, but making ing good progress, if this sounds like something you might want to use please sign up. | | |
| ▲ | dotancohen 3 months ago | parent [-] | | After debating for a few days, I joined the waitlist. My Gmail username is the same as my HN username. The website does a poor job of conveying what the product is exactly, which is the reason that I did not sign up sooner. You might want to approach the text of the site from the premise that the user has no idea what you are presenting. | | |
| ▲ | mehh 2 months ago | parent [-] | | Thanks for the feedback, and for joining the list. I agree with your points, my plan is to get a basic version in front of some people and shape it on their feedback. Plan to have it available in a couple of weeks. |
|
|
|
